A comparison of traditional and maximum likelihood approaches to estimating thermal indices for polymeric materials

Abstract

ABSTRACTAccelerated destructive degradation testing (ADDT) is a widely used technique for long-term material property evaluation. One area of application is in determining the thermal index (TI) of polymeric materials including thermoplastic, thermosetting, and elastomeric materials. There are two approaches to estimating a TI based on data collected from ADDT: the traditional and maximum likelihood (ML) procedures. The traditional procedure is specified in current industrial standards and is widely used in industrial applications. The ML procedure is frequently used in the statistical literature but rarely seen in industrial ADDT applications. At present, the ML procedure is not specified in the current standards. In this article, we compare both approaches with two motivating data sets from the literature and simulation studies. We show that the ML procedure has many advantages over the traditional procedure in terms of estimation performance, uncertainty quantification, material comparisons, and predictions. The comparisons and discussion in this article can be useful in designation of statistical methods for future industrial standards.